php-kilo/src/Tokens/PHP8.php

<?php declare(strict_types=1);

namespace Aviat\Kilo\Tokens;

use PhpToken;

use function Aviat\Kilo\tabs_to_spaces;
use const Aviat\Kilo\T_RAW;

class PHP8 extends PhpToken {
	private array $rawLines = [];

	private array $tokens = [];

	/**
	 * Use 'token_get_all' to get the tokens for a file,
	 * organized by row number
	 *
	 * @param string $code
	 * @return array
	 */
	public static function getTokens(string $code): array
	{
		// Make the lines/tokens 1-indexed
		$lines = explode("\n", $code);
		array_unshift($lines, '');
		unset($lines[0]);

		$self = (new self(0, $code));

		$self->rawLines = $lines;
		$self->tokens = array_fill(1, count($lines), []);

		return $self->organizeTokens($code);
	}

	/**
	 * Return tokens for the current $filename, organized
	 * by row number
	 *
	 * @param string $filename
	 * @return array
	 */
	public static function getFileTokens(string $filename): array
	{
		$code = @file_get_contents($filename);

		if ($code === FALSE)
		{
			return [];
		}

		return self::getTokens($code);
	}

	protected function organizeTokens(string $code): array
	{
		$rawTokens = self::tokenize($code);
		foreach ($rawTokens as $t)
		{
			$this->processObjectToken($t);
		}

		ksort($this->tokens);

		return $this->tokens;
	}

	protected function processObjectToken(\PhpToken $token): void
	{
		$currentLine = $token->line;
		$char = tabs_to_spaces($token->text);

		$current = [
			'type' => $token->id,
			'typeName' => $token->getTokenName(),
			'char' => $char,
			'line' => $currentLine,
		];

		// Single new line, or starts with a new line with other whitespace
		if (str_starts_with($char, "\n") && trim($char) === '')
		{
			$this->tokens[$currentLine][] = $current;

			return;
		}

		// Only return the first line of a multi-line token for this line array
		if (str_contains($char, "\n"))
		{
			$chars = explode("\n", $char);
			$current['original'] = [
				'string' => $char,
				'lines' => $chars,
			];
			$current['char'] = array_shift($chars);

			// Add new lines for additional newline characters
			$nextLine = $currentLine;
			foreach ($chars as $char)
			{
				$nextLine++;

				if ( ! empty($char))
				{
					$this->processStringToken($char, $nextLine);
				}
			}
		}

		$this->tokens[$currentLine][] = $current;
	}

	protected function processStringToken(string $token, int $startLine): void
	{
		$char = tabs_to_spaces($token);
		$lineNumber = $this->findCorrectLine($char, $startLine) ?? $startLine;

		// Simple characters, usually delimiters or single character operators
		$this->tokens[$lineNumber][] = [
			'type' => T_RAW,
			'typeName' => 'RAW',
			'char' => tabs_to_spaces($token),
			'line' => $lineNumber,
		];
	}

	private function findCorrectLine(string $search, int $rowOffset, int $searchLength = 5): ?int
	{
		$end = $rowOffset + $searchLength;
		if ($end > count($this->rawLines))
		{
			$end = count($this->rawLines);
		}

		for ($i = $rowOffset; $i < $end; $i++)
		{
			if (str_contains($this->rawLines[$i], $search))
			{
				return $i;
			}
		}

		return NULL;
	}
}